OVERVIEW
======================================
This directory contains maps between the raw survey data and the codes in `source/prepare_data` on the variable names and codings of the levels of categorical variables.

The folder `varname` contains a map at the year level for each country. Each column is a variable used in codes in `source/prepare_data`. The cell `(t, v)` records the variable name of the variable `v` used in the survey of year `t`.

The folder `party` contains a map between the party name and the variable name of the feeling on that party for each country. The map is at the year-variable level, where variable equals to either `parties` or `thermo_vars`. The level `parties` in the `variable` column records the coding of the levels of the variable `party_var` in maps in `varname`, which in most cases is the name of the parties. The variable `thermo_vars` records the variable names in the raw survey data of the feeling on the corresponding party. Values within the same column are always information about the same party. Notes:


The folder `thermo` contains a map between the thermometer and the thermo value, where thermometer means the wording in the raw survey and thermo value means the 100-scale value we converted to. The map is at the year-variable level, where variable equals to either `thermometer` or `thermo_vals`. Note that values within the same column are not necessarily at the same thermo value. So this map should not be used within columns and across rows. Instead, this map should be used within rows.

If same values apply for multiple years, all applicable years are recorded in the `year` column. Use comma to separate different years.


Notes on which party/affect questions to use:
 - Keep all parties for which at least one affect question is asked throughout the data series.
 - Keep all affect questions, even if no associated option for party affiliation. [These observations will be dropped in our main series, but will still be used in our data-driven series.]